Nutrient levels govern oceanic primary production, but measuring them is labor-intensive and costly. To address this, we used machine learning models to learn the hidden relationships between easy-to-measure ocean properties (like temperature and salinity) and nutrient levels. Applying this model, we created ~ 470 million nutrient data points across the North Pacific from 1895 to 2024. This data will help to understand nutrient dynamics and marine ecosystem variability under climate change.